Blend Labs, Inc. (BLND) Earnings Call Transcript
August 6, 2026
Earnings Call Speaker Segments
Good afternoon, and thank you for joining Blen's conference call covering the financial results for the second quarter of 2026. My name is Meg Nunnally, and I serve as Blen's head of investor relations. With me on the call are Nima Gamsari, our co-founder and head of Blen, along with Jason Rehm, who leads finance and administration for us. Prior to beginning today's discussion, please note that we'll be referencing some non-GAAP metrics, and you can find reconciliations of those to GAAP metrics in the earnings release issued today as well as in the appendix of our supplemental slides. Non-GAAP metrics shouldn't be viewed as a replacement for GAAP results. Unless stated otherwise, every financial metric we cover today, including our profitability, is gap. Additionally, some statements we make on this conference call concerning Blend and its operations — specifically our guidance for the third and fourth quarter of 2026, other commentary about 2026, and our expectations regarding markets, our strategic investments, product development plans, and operational targets — could be deemed forward-looking statements under federal securities law. We want to caution you that forward-looking statements carry substantial risks and uncertainties, and numerous factors, many outside the company's control, might cause actual results, events, or circumstances to differ materially from what those statements describe. Please review the risk factors we've identified in our most recent 10Qs, our 10K for the fiscal year 2025, and other SEC filings. We are not making any commitment to update these statements should conditions change, except as the law requires. The financial information shared on this call is based on continuing operations, and any prior periods have been recast to exclude operations that are now discontinued. Finally, a copy of our prepared remarks will be posted on our website by the time today's call concludes, and an audio replay will be made available soon after the call ends. I'll now hand the call over to Nima.
Thank you, Meg, and hello everybody. For Blend, the second quarter was once again a period of discipline and profitability. Our revenue landed close to the upper bound of the range we had guided, while non-GAAP operating income exceeded that upper bound. Jason will walk everyone through the complete financial picture shortly; for my part, I want to focus today on the two pillars underpinning our strategy. The first is Autopilot — those agents we create for our customers — and the second is Blend 3.0, the agents being built within Blend itself so that we can serve customers more quickly, with higher quality, and at lower cost. I'll begin with Autopilot. The major development from last quarter: Autopilot reached commercial availability on July 1st. If you're watching the webcast, please move to the slide titled our AI strategy, guiding our customers at the top. That's the first of two slides I'll point to today, and the charts appearing on them can also be found in the supplemental materials posted on our Investor Relations website. The chart tracks cumulative loans processed through Autopilot, starting from nothing in February and reaching beyond 45,000 as of today — and that number keeps compounding. This curve is the evidence behind everything I'm about to share. Across that four-month span, over 65 lenders switched Autopilot on. They put it through stress tests, brought the difficult edge cases to the surface, and influenced what we ultimately shipped. The preview data confirms why this matters; we're beginning to find proof that Autopilot produces quicker clearance times, better conversion rates, and possibly lower fulfillment costs. Drawing on our preliminary data for these loans that moved through our system, our customers are experiencing a 10 to 15 percent gain in pull through rates along with two to four days of improvement in cycle time. Beyond that, we estimate Autopilot handles 4.5 hours of loan fulfillment tasks per loan on average. That's enormous, and our customers have only started to appreciate what's possible. Since becoming commercial on July 1st, six lenders have already put pen to paper on contracts that include Autopilot, among them Onity, one of the country's biggest mortgage servicers, which is additionally constructing its own experiences on top of Autopilot via our Autopilot MCP server. On Moniton, we're carrying out the plan we laid out in May, with customers signing flat-fee one-year contracts for full access. A base level of intelligence will be baked into our workflows, but the complete product resides in the paid tiers; we refer to this as our underwriting intelligence, where Autopilot reads documents, performs calculations, reconciles against guidelines, and pushes the entire loan file forward. Eventually, our intention is still to shift the paid tiers toward a per funded loan model, exactly as with real estate, the rest of our mortgage suite; and when it comes to seed-based pricing, which in our view cannot survive the agentic world — when software performs the work, charging by the person makes no sense whatsoever. Plus, other companies are changing how they charge so they can earn money from the activity their AI produces. We took a different path: we only get paid when things go well. Payment comes from the result, not from each step taken to reach it, but from the result itself. So if autopilot performs 10 times as much work on a file that never closes, our customers should not have to pay 10 times as much, and with our approach, they will not. What we earn grows alongside what our customers achieve, and that connection has always been our business model. Agenda AI simply makes it more valuable and easier to scale. Because people constantly ask me about the competitive landscape, I will be straightforward. Why should Blend be the one to win? Our response boils down to four advantages that are difficult to copy. First, our position. Blend is the first place a borrower interacts with. And since we are present there, we notice the issues and resolve the issues as they appear. Over half of borrowers submit applications outside normal working hours, and more than 90% of those who finish an application do it within 24 hours of beginning. Autopilot detects friction right when it occurs, at 10pm on a Saturday, rather than Monday morning, precisely when borrower intent is at its peak and engagement is at its peak. Then there is data. The question underlying every AI question I receive is this: in a world where anyone can access a frontier model, such as Claude, can software like ours be replaced? Our view is no, and our data is why. Models are becoming more alike and everybody can use the same models, ourselves included. But what is not interchangeable is what it takes to make the model work on a mortgage that involves enormous context, enormous loan documents, guidelines, and many things that must be considered to determine the correct next action on the loan. And that comes from a mix of 15 years of experience and tens of millions of loan application data processing through our platform, which is not something you can purchase. And you cannot fabricate 15 years of actual borrower behavior; and that builds on itself. I genuinely observe this in our early autopilot benchmarks against a standard Claude plus skills. And we find that autopilot does better in those tests, and far better and far less expensive, roughly 1 third the cost. Third, the harness. Autopilot is not a wrapper around a generic model. It operates within the infrastructure. The data layer, the integrations, the compliance architecture, coordinating what the agent sees, what tools it may use, which guidelines apply to a particular loan in front of it, and what occurs when it is uncertain. It returns the file to the loan team. People remain in charge of the decisions that matter, and this harness is among the reasons Autopilot is more accurate and less costly than a generic harness; and last, our relationships. For 15 years, we have been building side by side with lenders, and Autopilot came together in that very same manner — alongside lenders, designed for lenders. Through Autopilot MCP, we have unlocked our infrastructure, enabling customers as well as other technology providers to build on top of us instead of working around us. Taking a step back, this represents the larger thesis. Blend is, and can be, the agentic infrastructure for relationship banking. Banking's initial promise centered on a relationship — a lender who came to know you and could make a decision based on more than merely a credit score. That promise did not vanish due to bankers ceasing to carry it. It vanished due to the sheer volume of process, alongside all the manual work and manual effort that must go into each and every loan. Yet when Autopilot takes care of that grunt work on the loans and assists the people who historically performed it in focusing on the customer, Autopilot is managing those conditions, the follow-ups, the questions arriving at two in the morning, and the loan officer can return to the capacity of serving the customer. Back in 2012, I founded Blend alongside my co-founders with a straightforward thesis: the mortgage process ought to drive itself. Not because humans aren't needed, but because the right technology can manage everything that does not require a human.
What we're seeing is that thesis for autopilot finally coming to pass. And given our observations in mortgage — the most complicated and most tightly regulated workflow in consumer finance — we think that same infrastructure naturally stretches into home equity, deposits, auto, card, and personal lending. And I'd say probably, more broadly, since we've sharpened the harness and the evals for autopilot so much, it extends past that into other areas of underwriting. Agentech AI isn't a substitute for relationship banking, nor does it have to be, yet it enables our customers — the lenders — to become even more centered on their own customers again, and to do so at scale. Changing topics, for those of you watching the webcast, I'd request that you move ahead to this slide, the one titled, our AI strategy transforming how we work. This slide offers a peek into how Blend itself is changing and what it will look like going forward. We refer to that as Blend 3.0. If you consider Blend 1.0 to be the initial 10 years, when we established a market leading company and expanded considerably — growing share, growing the customer base, and launching our first product — and Blend 2.0 to be the most recent four years, when we were building a profitable, long-term, sustainable entity, then Blend 3.0 is an agentic first company. That applies not only to our products, but also to how we operate internally. Last quarter I walked through this pattern: an agent will ultimately tell Blend to handle the initial pass of incoming work — particularly the grunt work I mentioned before — before any team member even touches it and brings their judgment to bear. So I won't go through all of that again today, but I'd like you to look at the graph. That graph shows one particular team inside the company — in fact one of our largest, the engineering team. If you look at the gray line, which represents pull requests in 2025, it stayed fairly flat all year and fell off during the holidays. Then if you look at the blue line for this year, we've multiplied our throughput as an engineering team by 3.6 with about the same headcount, just since January of this year. 3.6x the throughput. I'm not sure you recall, but back in May I told you we were running at about one and a half times what we'd been doing in January. That was the average — and that was only three months ago. Now we're at 3.6 times. That's a profound thing to contemplate, because it means if we keep up this pace, we could reach 10 times as much throughput on the engineering team by the end of this year compared with what we did at the end of last year. So I believe that's going to keep building on itself and work in our favor, letting us move faster when it comes to resolving issues customers report, taking care of their requests for features and improvements, and creating new offerings such as autopilot that drive where our industry is headed — doing all of that more quickly and more effectively. That matters a great deal, since what lies at the heart of every software business is the question of how you deliver for your customers and generate value on their behalf. And my sense is that this pattern will grow even steeper as we bring more AI into how we operate, and as those models improve as time goes on. Yet with the wins we're achieving in engineering, the following stage we're embarking on — and what I see as our responsibility — is bringing this across the entire company. So during the quarter just ended, we broadened our push around agent-first methods, extending into our go-to-market team, where agents now immediately examine incoming support tickets. When something turns out to be a bug requiring a fix, a pull request can be raised. When what's needed is a straightforward reply or a settings adjustment, the agent can prepare a draft so a person can simply click yes, that's correct, or no, that's not right, I need to alter it. Beyond that, it's handling tasks tied to our customer calls, assisting with drafting coaching notes and follow-ups that previously consumed hours of the team's time, and within our finance group, agents are completing the initial pass on portions of our closing work. Each instance follows an identical pattern: the agent produces the first draft, then a human checks and signs off. And in certain situations as time passes — where security concerns are lower, or code concerns are lower, or it's something that doesn't require a person's sign-off at all — my belief is that eventually we won't even need human approval for a few of those items. These are separate functions I'm describing, yet it amounts to a single operating model reflecting how I see the future of a GenTech technology. We're still in the early stages, but this ranks among the hottest subjects within Blend at the moment. Our leadership group gathers on a routine basis to discuss it, since turning into an agent-first company amounts to an operating choice rather than some peripheral initiative. It's an all-hands undertaking to determine how we might rethink this remarkable company and its customer base starting from scratch, because that chance is now available to us thanks to our profitability. We've gotten our house in order, and all of these technologies are speeding forward before our very eyes; back in May I stated that our aim is to rank within the top 1% of all companies when it comes to adopting agentic AI, and that remains our target. We're monitoring how we're doing function by function. We're proceeding one step at a time, and I anticipate telling you more in the quarters ahead as this rollout reaches maturity. Finally, I'd like to address growth, since I realize it's something everybody's wondering about. Let me begin with what isn't good: current macroeconomic conditions aren't doing us any favors. Due to conflicts and other developments beyond our influence, mortgage rates climbed from roughly 6.4% in May to 6.8% over the past few weeks. That's holding back market activity — refinancing in particular, though purchase activity too. As I've consistently noted, the macro isn't something we have any control over. My reason for flagging it is simply that we're clearly monitoring it closely, and it could dampen upcoming periods such as Q4 — if we were counting on a particular level of refinance volume and it fails to materialize due to elevated rates, that's just something to keep in mind. Still, that's fine. Generating long-term sustainable growth no matter what the macro environment looks like has always been our ultimate objective, and it remains so. During the quarter, we signed 14 new deals and expansions, and two of them deserve a mention. The first is a new logo involving a sizable credit union that brought in Autopilot from the very start. I believe this will become standard practice for our customers when they sign with us. We have additional deals in the pipeline that look similar, but this marks the first instance of a customer picking Blend and picking Autopilot within that initial package.
Agentic AI has become something customers take for granted — they want it, and honestly they need it, in order to perform at their absolute best. As I said earlier, I think it shows us what a big chunk of our deals will look like in the future. The next deal I want to talk about is one where we sold RapidRefi together with RapidHomeEquity into a top five credit union — they signed on with us a couple of years ago, the rollout took place, they're happy, and it's a great example of growing and deepening relationships with clients as we bring those initial projects through to success. At the same time, our pipeline continues to grow. On the last call, I noted that our total pipeline was up more than 40% year over year, and that total pipeline is still growing. What gets me more excited as the year goes on is a narrower segment. We brought in new sales leadership at the start of this year, and our late-stage pipeline — the deals we plan to close within the next quarter — grew nearly 40% from March to June by itself. Late-stage pipeline is what's truly near signing, so it comes with much better visibility and confidence. To give you a sense of that pipeline, it includes another large mortgage customer, a top 20 financial institution, and a solid group of rapid and autopilot deals — for example, we expect two more large rapid deals to close in the coming months. Now, I want to be straightforward about timing. We expect our pipeline to turn into signed deals within roughly the next quarter, though some may slip. Those signed deals then become revenue. But all of this involves some of the largest financial institutions in the country, so when they sign and go live, we expect it to show up in our financials over the medium term — yet very large financial institutions need time to navigate their governance processes for things like agentic tools. So with all that said, the direction is pretty clear. Our customers are excited about it. Our biggest customers are really leaning in, and my confidence is very high. The sales discipline Matt has put in place isn't limited to new business alone. It also affects how we manage renewals, and non-GAAP operating income came in $7 million above the top end of our $5.5 million to $6.5 million guidance range. We're doing a better job of bringing customers into renewal discussions sooner, which means more relationship components in how we work with them, renewing customers for longer terms, expanding the relationship at renewal, and making sure our pricing reflects the value we deliver — Jason will give more detail on that later, on how we're handling this, where renewals and upsells are coming from, and where we're seeing churn. Okay, let me tackle that other side right now, because fundamentally, our existing customers are renewing for longer terms and larger amounts. In summary, the market over the near term, together with macroeconomic circumstances, is largely what it is, but we stay disciplined and profitable in that environment. Our focus is on the medium and long term — that is, 2027 and beyond. As autopilot becomes commercially available and scales, given the pipeline we're building and the momentum we're gaining from the agentic transformation taking place within the company, I'm more energized than I could possibly be about where Blend sits when this cycle turns. That applies not just to us, but to our customers as well. And ultimately, because our pricing approach is value-based, that determines what it means for our financial results and for our investors. Having said that, I'll pass it to Jason to walk through the financials. Thanks, Nima, and thank you to everyone else joining us on the call. We delivered a solid second quarter, with total revenue of $33.8 million, up 7% year over year and near the top of our guidance range of $32 million to $34 million. Mortgage Street Revenue was $19.2 million, a 7% year-over-year rise, within the growth range we discussed on our prior call. Funded mortgage loans via our platform came to about 233,000 in Q2, up 14% year over year and consistent with what we anticipated going into the quarter. That volume growth was partially offset by a lower economic value per funded loan versus a year ago, coming in at $79, which is consistent with the $79 to $80 range we guided to in May. As a reminder, the decline from $83 in Q1 is largely arithmetic. Q1 is seasonally the high point because mortgage volumes are lower then, and the greater volumes in Q2 naturally bring down the per-loan figure, since some fee arrangements with our customers are fixed. Consumer banking suite revenue for the second quarter was $12.2 million, up 6% year over year, slightly above the top of the growth range we shared on our earlier call. And professional services revenue for the second quarter came to $2.4 million, in line with our expectations. Turning to profitability, non-GAAP gross profit was $26.5 million, and our non-GAAP gross margin was 78.3%, up from 76.1% in the second quarter of 2025, and consistent with the normalized gross margin framework of 77 to 78% we set out last quarter, after excluding the one-time benefits from Q1. That gross margin improvement occurred even with the model costs related to autopilot, which remained fairly low but were rising in Q2. We expect those costs to remain relatively immaterial to our P&L in the near term, though we'll update you if anything changes there. Non-GAAP operating expenses were $19.5 million in Q2, essentially flat year over year. Non-GAAP operating income exceeded the high end of our $5.5 million to $6.5 million guidance range by $7 million. Non-GAAP operating income landed at $7 million. That sits above the high end of the $5.5 million to $6.5 million range we had guided to — $7 million, in fact, exceeded that top end. It translated into a non-GAAP operating margin of 20.6%. That figure improved by 2025. Free cash flow for the quarter was $6.9 million. As the quarter ended, we were holding $44.9 million across cash, cash equivalents, and marketable securities, and we still have zero debt on our books. During Q2, we repurchased 11 million shares at an average price of $1.65 each. Year to date, 22.2 million shares have been bought back for $36.8 million, leaving about $13.2 million remaining under our newest authorization to repurchase $50 million of shares. Our thinking hasn't changed: share repurchases remain an excellent use of capital, especially considering the current valuation, though we'll weigh any future buybacks against our wish to maintain ample liquidity for running the business. Before diving into details, let me frame how we're reading the environment, because it shapes everything I'm about to discuss. During our May call, consistent with what we've said before, we noted that our own outlook for the second half of 2026 was nearly identical to Fannie Mae's, which at the time projected mortgage market volume growth of about 19% for the full year. Fannie has now lowered that estimate to roughly 17%.
At this moment, our view is somewhat more restrained than what Fannie Mae is projecting. We expect refinance volume to remain weak as long as rates stay high for a prolonged period. To put a finer point on it, Fannie's expectation of refi expansion in the back half of the year is difficult for us to get behind, especially with rates where they are. Fannie Mae will update its forecasts, but until then, we are taking the more careful position. Beyond that, churn notices came in higher this quarter relative to recent quarters. These notices reflect customers signaling an intent to exit Blend, although the timing and ultimate significance of that remains uncertain. In most of those cases, the plan is to move toward lower-cost or zero-cost point solutions. We think the revenue impact can be absorbed, amounting to the low single digits of yearly revenue, but we are watching it closely, and as a result, we have incorporated some conservatism into how we think about the back half of the year. With that context, let me walk you through what we expect over the next 2 quarters. Starting with Q3, total revenue ought to come in at $31.5 to $33.5 million, representing approximately negative 4 to positive 2% growth year over year. Underneath those headline numbers, we assume a total mortgage market of 1,200,000 to 1,260,000 units, which is up 5% at the midpoint, plus Blend funded loan volume of about 200,000 to 210,000 loans, up 2% at the midpoint; that translates into a mortgage suite revenue change of roughly negative 4 to positive 3% year over year. On economic value per funded loan, we estimate approximately $80 to $81. We anticipate consumer banking suite revenue growth year over year of negative 5 to positive 1% for Q3, consistent with the moderation we described on our prior call; and for Q3 non-GAAP operating income, we expect a range of $3.5 to $4.5 million, implying a non-GAAP operating margin near 12% at the midpoint. Bear in mind that Q3 expenses carry our annual Blend Forum customer event, as they have in prior years, and that should drive a sequential increase of about $1.5 million in sales and marketing expense quarter over quarter. Beyond Q3, I want to offer our current view on fourth quarter volumes. Fannie Mae's newest forecast indicates market volume that is approximately flat in Q4. As I mentioned earlier, that projection may be revised at some point, but for now, our stance is a bit more cautious, and we expect a total market size of 1.105 million to 1.165 million units, or a decrease of roughly 11% year over year at the midpoint. Separately, for Q3 non-GAAP operating income, we expect $3.5 million to $4.5 million, and we'd expect Blend's fourth quarter funded loan volume to be approximately 180,000 to 190,000 loans. So with all of that in mind, and factoring in the last bit of runoff from that big customer we've discussed on prior calls, our expectation for Blend's Q4 funded loan volume is roughly 180,000 to 190,000 loans, which works out to a decline of about 10% to 15% compared with a year earlier. That's our best read right now, but I want to flag that the macro picture remains quite fluid and very sensitive to rates. All right, let me close on autopilot — you've already heard Nima's comments on this, and the early commercial traction has us truly excited, though we'd continue to caution against building autopilot revenue into your models at this point. Later on, after customers and prospects have moved through the funnel and we've had a longer window to watch, we plan to provide more detail on what this might mean for our model. In summary, this was the quarter we were aiming for. Revenue landed near the top of the range, profitability came in above the high end, and we carried out a strategic return of capital through our share buyback. Macro conditions continue to work against us, and we've tried to give you a clear view of how we see market volumes. Even so, the fundamental levers we control — customer wins, product velocity, and now autopilot's commercialization — are all moving in a favorable direction. Those factors put us in position to reaccelerate growth as we head toward 2027. Oh, great. Thank you. Nima, last quarter you said you thought agents and autopilot could be a path to potentially delivering 10% to 15% incremental revenue growth in 27. Now that you're a month into selling the one-year flat rate autopilot contracts, with six lenders signed up, and given your comments that governance at large financial institutions takes a little while to navigate, does that view still hold today? Yes, I don't want to give any additional guidance beyond that, but yes, the pipeline looks good both in terms of what we sign, and I'd add that a substantial portion of our biggest finance — I think what surprised me both last quarter and this one was just how aggressively our largest financial institutions are leaning in. Let me share one anecdote: at some of the biggest banks in the country, this has always been an executive priority, but watching the business line and the tech lines at these organizations push the rest of their company to make this happen has genuinely surprised me. It takes a lot to stand up and say, we really need to make this happen for our business, yet that's exactly what they're doing. I believe that's why our largest customers are leaning in. Some are either in late-stage trials with us or in final approvals with their internal committees. But as I mentioned, these things take time and we're very excited about it. On top of that, for the third quarter we expect non-GAAP operating income in the range of $3.5 million to $4.5 million. Also, and this might matter even more — I brought this up quickly during the prepared remarks — but those first figures coming out of autopilot, wow, we've seen about 50,000 loans move through it during the past few months. That lets us see some of the gaps between pre- and post-numbers for those lenders. Watching it is truly thrilling, especially since we're at such an early stage and two favorable forces are at play for us: our own work picking up pace and the models getting stronger. Our efforts on the harness, honestly, are speeding up, and the models underneath keep improving as well. We've set things up so that this turns into a compounding engine for us. And like I said, it's early — as I mentioned, it's early — yet the first signals look better than I would have accepted had someone told me in January of this year that we'd build this product from scratch and this is where we'd be 6 months later. Understood. Yeah, that chart on slide 8 with cumulative loans processed every 2 weeks strikes me as quite persuasive. Then my second question concerns Blend 3.0 — how do you balance the trade-off between innovating faster and shipping better products against achieving OpEx leverage? Honestly, that's one of the top things investors ask me. Based on today's update, when does that 3.6 times increase in pull requests convert into either revenue or operating leverage? Excellent question. I believe part of the revenue story is already covered by what I mentioned earlier — in a market as heavily regulated as this one, these products take time to materialize. Looking ahead, we expect Q3 non-GAAP operating income to land between $3.5 million and $4.5 million, and for Blend's fourth quarter funded loan volume, we'd expect roughly 180,000 to 190,000 loans.
It's really a component of our competitive advantage, honestly, that we've got these strong ties with clients who place their trust in us and grant us the first opportunity to tackle their toughest challenges — in this instance, that means their operations around how they can make underwriting and processing mortgages superior, quicker, and more economical. But it does require time. That's the reason I've continued to emphasize — I directed our internal team toward the medium term, meaning 2027. Once these agreements progress, once they're up and running and reach scale, how can we turn that into an advantage for ourselves, rather than concentrating on generating instant results today at the cost of potentially sacrificing 2027 and 2028 outcomes? And that's what's wonderful about being a profitable business at present in this awful mortgage landscape with macroeconomic conditions pushing against us — we're able to keep our eyes on the medium and long term, which I'm aware our investors deeply value. For instance, what will this organization become, you know, with an agentic core offering fueling much of the expansion and with a macro environment that revives, what will it become under those circumstances? So that's where our attention is firmly placed. I take pride in how the team is performing right now. Clearly, there's considerably more effort ahead for us. And simply the degree of dedication and ferocity the team has brought to these efforts has been wonderful for me to witness, as somebody who has clearly been present since the very beginning. Good afternoon. Appreciate you taking the question. To start, Neema, it's wonderful to observe the effect Autopilot is delivering on the mortgage side. I'm curious, is Autopilot likewise offered to your consumer banking customers, and perhaps you could address the potential there if there's anything new to share, and then I'll have a follow-up. Yes, it's incorporated into our platform, and so, you know, we possess the foundational elements required to serve all of our clients, which they're enthusiastic about. It's a significant ask we receive from our clients — how can you assist some of them who have several product lines with us? How can you assist this other business line that continues to manually confirm IDs or documents that are completed and forwarded to them? So we haven't launched that yet, but I'd suggest keeping an eye out. It's something we're giving serious thought to and we've had some initial client involvement around. And just to offer a bit of insight into our method for developing products such as this — choose a handful of Lighthouse customers, go deep with them, demonstrate the value, and afterward extend it to the broader market soon after. And I'm not referring to a span of a year. I'm referring to a span of a quarter or two, and push that forward as rapidly as we can. Understood. And you brought up two deals within the credit union channel. Perhaps we could hear how that channel is doing overall and what the pipeline looks like there. Thank you. Yes, we don't break out credit unions separately from banks, independent mortgage companies, or other financial institutions. Even so, here's what I can share: the customers we work with are genuinely robust, spanning everything from the largest credit unions down through mid-sized and community ones. Within the groups we cater to, my sense is that these are the institutions prioritizing self-service for their members above all else. So they're a great match for how we think about our products — enabling someone who submits an application at 2 in the morning to walk away with a complete approval. The rapid suite we've been talking about demonstrates this perfectly, and so does autopilot, especially when it comes to getting a person an initial offer followed by a full approval overnight. That, I think, explains why our interests line up so well with what this segment is after — the goal for them is just getting as many members signed up as they can. The self-serve member experience is something they like, and it fits nicely with where we see our product going over the long term. Hello everyone, Juan Chong here, stepping in for Ryan. Appreciate you all for the questions. I'd like to go a bit further into what Jason mentioned regarding the rise in chart notices we saw this quarter. I understand the effect there is pretty minor, but could you expand on it a little? Would these be things like the AI native point solutions that have just hit the market lately, or is it maybe that customers are increasingly choosing to build in house? No, that's a fair question. Honestly, what we observed was customers moving over to those cheaper options already out there, though I wouldn't go so far as to label that a pattern yet. There simply aren't sufficient data points at this stage. That said, here's my take: sure, this rate climate plus the broader economy is hurting us somewhat when it comes to mortgage volume, but honestly our customers are feeling that pain just as badly, if not worse. They're hunting for any way to cut costs, and things are genuinely difficult on their end too. So in the handful of instances we've encountered, that's likely what's behind it. Got it — that makes sense. One more question from me. A bigger shift we've been watching in the mortgage space overall involves share moving in the direction of the broker channel. What's your take on how this might shape blend, and which direction do you see the trends heading? Yeah, what we've seen is something that's been building for several years now — it isn't just a blip. In the broker channel, it climbed from around single digits to roughly 15%, and in some cases reached the upper teens. That said, here's my read: our customers — whether they're independent mortgage banks, banks, or credit unions, regardless of size — a good number of them are really committing to this market. There's something I keep hearing, and it ties back to what came up in the previous question as well: the companies that buy from us are placing their bets on what's ahead. They're wagering that by adopting these emerging technologies and by getting the fullest possible value out of our platform — and out of the other platforms they run — they'll continue to expand. So what I'm noticing, particularly in this year, is that maybe during 2025 and 2024 folks were tidying up their own affairs, much the way we might have been doing ourselves. Yet in this year above all, we're watching businesses within our client roster declare that now is the moment to put money into what comes next. My sense is that this likely contributed to some of the movement toward different channels, since certain companies weren't in a position to spend during 2023, 2024, or a portion of 2025. Still, with the market having settled, I believe our clients are putting capital toward the future. So I can't make claims about the entire market, though what I can say about the customers we serve is that they're genuinely committing. They're after fresh solutions. They're after serving their own customers more effectively. They're after stronger internal operations.
Given how much emphasis our customers place on getting better, I believe that base of customers we serve will keep coming out ahead in this market.
Hey Nima, Jason, Meg, thanks a lot. Nima, I've got a question for you about autopilot — the early interest is encouraging, and you shared some solid stats on the improvement and the reduction in pull-through and cycle time, and honestly each one of those stands on its own as compelling. But I'm curious how you're viewing the chance to sort of coordinate the whole funnel of the workflow, and how all those separate pieces stacking together feed into customer ROI — maybe how that builds conviction around the autopilot strategy, and really what sets it apart from some of the possible alternative tools or capabilities out there.
Yeah, and let me add a bit of color, since I think this is a really deep question — just to paint a picture of how typical solutions in the market look to me, and I keep seeing something new appear, whether that's from an incumbent back office or middle office company, or even from other folks in our space, and much of what they do is take a process that already exists and put something new on top of it. So, say, you've got a document that got uploaded, and they add a button or a little text entry field on that document asking, hey, did this document meet the guidelines? What that's really saying is we believe the existing process needs more intelligence baked in, which is precisely the opposite of how I see this concept I've now discussed for two quarters — the background worker agency, the ambient intelligence I mentioned in prior quarters. In fact, it's the antithesis of autopilot. So I think you're exactly right with your specific point about how we orchestrate the whole process. And much of what I've worked on with our team regarding the vision for autopilot, and especially joining forces with Rapid, where Rapid delivers you a real time offer and then autopilot ought to take you in that first session from that real time offer through to clear to close — that's where the market has to head, where a consumer can come in and have a self driving process where they receive a great personalized offer and they're ready to close. And all of that happened at two in the morning. And that's all achievable today, no matter what existing infrastructure is out there. It's kind of, you know, the beauty of autopilot being an open harness able to hook into our system or any other system is that it doesn't care which systems are in place. It simply says, here's the work that has to get done. Let's get it done right now. And it's not there yet when it comes to taking a customer, you know, from an offer with Rapid all the way through clear to close in autopilot. Yet that’s where things are pointed, and we’ll keep at it right up until loans can be signed off on at two in the morning. And with the platform reviewing each piece of paperwork, asking for additional details, and engaging with the borrower live to obtain what’s needed — that, I mean, is where intelligence is headed. In a marketplace like this one, there’s simply no justification for anything less. Think about a credit card: it would be baffling to sit around for forty-eight hours, or two weeks, before learning whether your application got the green light. And it makes no sense that in this marketplace, given that this technology is now available, things would still work that way. I say this having spent fourteen years in this marketplace and genuinely grasping how deep and complicated it is to work through thousands of pages of rules plus hundreds of pages of loan paperwork, and how the autopilot harness is designed to handle that through a contextual, step-by-step underwrite — or underwriting helper, so to speak. That, I mean, is the sort of capability I believe is within reach today. That’s the goal we’re pushing toward. And that’s where this compounding advantage — to borrow your phrasing — will kick in for our clients, who’ll manage to accomplish things at prices they never imagined could exist.
Understood. Really useful. Thanks, Nima. And perhaps one brief question for Jason. I appreciate the detail on how the outlook works mechanically — if we look at EVPFL, which is being brought into queue and is expected to rise somewhat, I suppose I just want clarity on how much of that comes from the mechanics of, say, a smaller volume boost as opposed to possible added cross-selling of products. And I realize we aren’t claiming autopilot is contributing at any level, but
whether to any degree it’s also fueling part of that increase. Thanks. Yeah. I mean, our expectations for volume aren’t hugely different between Q2 and Q3, so I believe it’s largely just the business dynamics. Clearly we aren’t steering toward a notably different figure. It isn’t a major shift in either direction; so I’d regard Q2 and Q3 as roughly steady state relative to one another.
We aren't taking any more questions right now. That brings an end to the call for today. We appreciate your participation. You're free to disconnect at this point. This real-time transcript is produced automatically, with no person involved and no review.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Blend Labs, Inc. transcript - plus 131,000+ transcripts from 7,000+ companies, speaker segments and full-text search - through the EarningsAPI REST API or hosted MCP server.
Get an API key View API docs →For developers and AI pipelines
Programmatic access to Blend Labs, Inc. earnings transcripts and 131,000+ others is available through the
EarningsAPI REST API and the hosted MCP server.
Quarterly plans from $162 - full transcripts, speaker segments, full-text search,
and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.